The proposed randomized controlled pilot study will inform the design of a fully powered trial that will test the global effectiveness of a nurse-led interdisciplinary care management intervention for improving the health profile of uninsured diabetic patients who receive their primary medical care from a community-based free clinic compared to similar patients who undergo peer-led self management. The comparative effectiveness of these approaches has not been tested in a randomized trial. Thirty patients will be randomized to the care management intervention for the first 6 months. Major components of the nurse-led care management model include: 1) patient-centered assessment and goal setting to maximize patients' self-management skills; 2) education to improve self efficacy and promote productive clinical encounters; 3) access to specialist care through an interdisciplinary team; 4) delivery of customized evidence-based treatment recommendations in response to patients' goals; and 5) care coordination by a nurse care manager. For the second 6 months patients will return to usual clinic care. The peer-led self management group will be invited to attend weekly peer-led self-management training following the Stanford Train the Trainer Model, a strictly proscribed program ensuring uniform application. The training will occur for 2 hours per night for 6 weeks and will cover self-management of chronic conditions, making action plans, feedback and problem solving, handling emotions, symptom management techniques, communication with health care providers, medications and making treatment decisions. The peer advisors will then lead, with assistance from a nurse, 1 hour monthly group sessions for the remainder of the first 6 months for review/revision of action plans, review of self management techniques and ongoing social persuasion and modeling. For the second 6 months patients will return to usual clinic care. In addition to demonstrating feasibility, refining inclusion criteria, and providing data for sample size determination, this pilot will help us select appropriate measurement tools to define a comprehensive profile of health and wellbeing to be used for comparing the groups and for estimating change within each group. The profile of measures will be obtained at six and 12 months and fall into five domains: 1)Medical risk management, hospitalization days, quality of life, quality of medical management, and quality of self- management. For this pilot, we will not test any hypotheses since protecting Type I error for such a large number of tests would yield inadequate power for any reasonable sample size. Instead, effect sizes and confidence intervals will be calculated for each domain as well as the overall test, and these results will be used for power analysis. In the fully powered trial, comparison of the groups across the 5 domains will be accomplished using an innovative global hypothesis testing strategy for multiple endpoints used in our other federally funded chronic illness management trials. A cost effectiveness analysis will also be performed.